Submitted 15 September 2016Accepted 19 December 2016Published 25 January 2017

Corresponding authorEdmund Februaryedmundfebruaryuctacza

Academic editorCoen Ritsema

Additional Information andDeclarations can be found onpage 13

DOI 107717peerj2923

Copyright2017 Shadwell and February

Distributed underCreative Commons CC-BY 40

OPEN ACCESS

Effects of groundwater abstraction ontwo keystone tree species in an aridsavanna national parkEleanor Shadwell and Edmund February

Department of Biological Sciences University of Cape Town Private Bag Rondebosch South AfricaThese authors contributed equally to this work

ABSTRACTBackground In arid systems with no surface water deep boreholes in ephemeralriver beds provide for humans and animals With continually increasing infrastructuredevelopment for tourism in arid wildlife parks such as the Kgalagadi Transfrontier Parkin southern Africa we ask what effects increased abstraction may have on large treesLarge trees in arid savannas perform essential ecosystem services by providing foodshade nesting sites and increased nutrients for many other plant and animal speciesand for this are regarded as keystone speciesMethods We determine seasonal fluctuations in the water table while also determiningthe water source for the dominant large tree species in the Auob and Nossob rivers inthe Park We also determine the extent to which these trees are physiologically stressedusing leaf δ13C xylem pressure potentials specific leaf area and an estimate of canopydeath We do this both upstream and downstream of a low water use borehole in theAuob River and a high water use borehole in the Nossob RiverResults Our results show that the trees are indeed using deep groundwater in the wetseason and that this is the same water used by people In the dry season trees in theAuob downstream of the active borehole become detached from the aquifer and usemore isotopically enriched soil water In the Nossob in the dry season all trees useisotopically enriched soil water and downstream of the active borehole use stomatalregulation to maintain leaf water potentials These results suggest that trees in the moreheavily utilised Nossob are under more water stress than those trees in the Auob butthat trees in both rivers demonstrate physiological adaptation to the changes in availablewater with smaller heavier leaves no significant canopy dieback and in the dry seasonin the Nossob stomatal regulation of leaf water potentialsDiscussion An increase in abstraction of groundwater particularly at the Nossob bore-hole may cause an additional draw down of the water table adding to the physiologicalstress demonstrated in our study The managers of the Kgalagadi Transfrontier Parkhave a mandate that includes biodiversity conservation To fulfil this mandate upperand lower thresholds for groundwater abstraction that allow for an adequate ecologicalreserve have to be determined

Subjects Biodiversity Conservation Biology Ecology Ecosystem Science Soil ScienceKeywords Acacia erioloba Acacia haematoxylon Water use Water stress Physiological stressCanopy dieback Tourism Threshold for potential concern

How to cite this article Shadwell and February (2017) Effects of groundwater abstraction on two keystone tree species in an arid savannanational park PeerJ 5e2923 DOI 107717peerj2923

INTRODUCTIONAll of the perennial rivers in the Kruger National Park in South Africa have been anthro-pogenically modified by water abstraction and increased sediment and pollution loads (DuPreez amp Steyn 1992 Rogers amp Biggs 1999) It has also been demonstrated that ephemeralrivers such as the Auob and the Nossob in the more arid Kgalagadi Transfrontier Park arecurrently not threatened by water abstraction even though the first boreholes were sunk inthese rivers in the 1930rsquos (Mills amp Retief 1984 Nel et al 2007 Van Wyk amp Le Riche 1984)Tourism development is seen as fundamental to community development and poverty alle-viation inmany parts of theworld and is therefore heavily promoted inmanyNational Parks(Binns amp Nel 2002Neto 2003) To this end there have been substantial increases in touristinfrastructure atNossob Kieliekrankie andUrikaruus tourist camps in theKgalagadi Trans-frontier Park (SANParks 2016) (Fig 1) As there is no natural surface water in this park allthe water for this increase in tourism development has to come from groundwater abstrac-tion (Mills amp Retief 1984 Van Wyk amp Le Riche 1984) The economic benefits of tourismhave been illustrated in several publications while the environmental impacts resultingfrom groundwater abstraction for this have also been demonstrated (Hall 2001 Mbaiwa2003 Stromberg 1993) Studies have shown that an increase in infrastructure developmentin arid environments can result in both short and long term water table declines withnegative repercussions on groundwater dependant ecosystems (Barron et al 2014 GroomFroend amp Mattiske 2000 Lite amp Stromberg 2005) The effects of such abstraction on vege-tation structure in arid systems with no surface water have however not been demonstrated

Arid savanna such as at our study site in the Kgalagadi Transfrontier Park occuracross the world from the Caatinga in northeast Brazil through to the Mulga of centralAustralia (Huntley amp Walker 2012 Nix amp Austin 1973) In arid savanna trees are scatteredin an expansive bare ground shrub and grass matrix At our study site Acacia erioloba andAcacia haematoxylon dominate in the dry river beds as the only large tree species along withscattered smaller trees and bushes with dwarf shrubs and grasses occurring in sparse clumps(Leistner 1959 Van Rooyen et al 2008) Both A erioloba and A haematoxylon providenesting sites shade food resources and soil nutrients for a variety of other plants andanimals and because of this are considered as keystone species (Dean Milton amp Jeltsch1999Milton amp Dean 1995)

Recent research has demonstrated that trees growing in the Kuruman River a systemsimilar to the Auob and the Nossob are using deep (56 m) water (Schachtschneider ampFebruary 2013) If these trees are indeed reliant on deep water and this is the same waterthat is being pumped for tourism development then there is an urgent need to develop aconceptual understanding of groundwater recharge processes and the interaction betweengroundwater and the trees growing in the river bed Several studies have now demonstratedthat it is possible to determine the water source of a plant through an analysis of the stablehydrogen and oxygen isotope ratios of the water in the xylem tissue of that plant (Dawsonamp Ehleringer 1991 Ehleringer amp Dawson 1992 February et al 2007) The method is basedon the assumption that the water in the xylem of non-photosynthesising tissue willhave the same stable isotope ratio as the water source (Dawson amp Ehleringer 1991) We

Shadwell and February (2017) PeerJ DOI 107717peerj2923 217

BOTSWANA

SOUTH AFRICA

Urikaruus W

Mata Mata

Nossob

Nossob

Riv

er

Nossob

Riv

er

Auob

River

South Africa

Twee Rivieren

N

0 10 20 30 40 50 km

Kieliekrankie

North Kwang

Qubitrsquoje Quap

Borehole

Rest Camp

Kamqua

Urikaruus

Figure 1 Map of the Kgalagadi Transfrontier Park showing the location of the four boreholes at whichboth water and twig samples were collected The borehole for the Nossob Rest Camp is between theNorth Kwang and Qubitrsquoje Quap boreholes with a pipeline down to Nossob

determine the interaction between groundwater and the two tree species A erioloba andA haematoxylon growing in the dry beds of the ephemeral Auob and Nossob rivers in theKgalagadi Transfrontier Park (Fig 1) We do this by firstly determining the water sourcefor the two species using hydrogen and oxygen isotope ratios of both xylem and boreholewater (Schachtschneider amp February 2013) We then determine the extent to which thetrees are physiologically stressed using stable carbon isotope ratios of the leaves middayxylem pressure potentials specific leaf area and an estimate of canopy death (Farquhar etal 1989 Liu amp Stuumltzel 2004 Scholander et al 1965)

Shadwell and February (2017) PeerJ DOI 107717peerj2923 317

National Parks in South Africa have adopted strategic adaptive management with clearecosystem management goals The achievement of these goals is evaluated by usingenvironmental indicators (Biggs amp Rogers 2003) With a constant demand for resources inthe Kgalagadi Transfrontier Park there is a real need to ascertain what the natural balanceof groundwater is and what the interaction is between this water and the trees growing inthe dry river beds Such an understanding is critical for determining the influence of watertable flux on riparian species thereby allowing park management to develop a sustainablewater resource allocation that meets the needs of both tourism and the environment (Chenet al 2016)

METHODSStudy siteThe study site is located in the western and southern part of the Kgalagadi TransfrontierPark in southernAfricaWe sampled at four boreholes in the ParkUrikaruuswaterhole andKamqua waterhole upstream and downstream respectively of the Urikaruus Wildernesscamp borehole (minus26010832 20349627) in the Auob River and North Kwang andQubitrsquoje Quap boreholes upstream and downstream respectively of the Nossob campborehole (minus25287383 20537513) in the Nossob River (Fig 1)

The climate for the region is characterised by distinctly seasonal rainfall with a hot wetseason from late November to early April Mean annual rainfall (1984ndash2014) increases from180 mm at Nossob (minus254212 205968) in the north to 220 mm at Twee Rivieren 116 kmto the south (minus264721 206116) Mean annual maximum and minimum temperaturesat Twee Rivieren are 367 C and 01 C for January and July respectively (South AfricanWeather Bureau)

The general landscape is primarily covered in aeolian dune sandunderlain by silcretes andcalcretes of the Cenozoic Kalahari Group Our sampling sites in the riverbed however typ-ically consist of fine grained silts (Mucina amp Rutherford 2006) In the Auob A erioloba andA haematoxylon (biogeographically endemic to the Kalahari) are the dominant tree specieswith Acacia mellifera and Rhigozum trichotomum common shrubs Grasses such as the an-nual Schmidtia kalahariensis and perennial Stipagrostis obtusa occur in sparse clumps (Leist-ner 1959Mucina amp Rutherford 2006) The vegetation in the Nossob is very similar to thatof the Auob but without A haematoxylon only large A erioloba trees Common grasseshere are the perennial Panicum coloratum var coloratum and Eragrostis bicolor (Bothma ampDe Graaff 1973Mucina amp Rutherford 2006)

Water sourceBorehole water levelTo monitor fluxes in water table through time piezometers (Solinst-Leverlogger George-town Ontorio Canada) were inserted into our sampled boreholes North Kwang andQubitrsquoje Quap in the Nossob and Urikaruus and Kamqua in the Auob from the 25thAugust 2012 to the 31st January 2014 (Fig 1)

Shadwell and February (2017) PeerJ DOI 107717peerj2923 417

Oxygen and hydrogen stable isotopesWe use the hydrogen and oxygen stable isotope ratios of water extracted from non-photosynthesising xylem tissue to determine the water source for both A erioloba andA haematoxylon (February et al 2007 Schachtschneider amp February 2013) The method isbased on the assumption that water extracted from non-photosynthesising xylem tissueof a tree will have the same isotopic ratio as the source water for that tree (White Cookamp Lawrence 1985) For this we collected two twig samples (sim05 cm times 6 cm) from sixA erioloba and six A haematoxylon (twelve trees) upstream and six trees for each species(twelve trees) downstream of the active borehole for the Urikaruus Wilderness Camp (Fig1) At the same timewe also collected two twig samples from eightA erioloba trees upstreamand eight trees downstream of the active borehole for the Nossob camp (Fig 1) Thesesamples were collected into borosilicate tubes (Kimax-Kimble New Jersey USA) whichwere placed directly onto a cryogenic vacuum extraction line to separate out the water forstable isotope analysis (Schachtschneider amp February 2013) We did this on three occasionsin January 2013 (wet season) July 2013 (dry season) and January 2014 (wet season)

We also collected water from each of our four boreholes at Urikaruus Kamqua NorthKwang and Qubitrsquoje Quap in August 2012 (dry season) November 2012 and January 2013(wet season) July and August 2013 (dry season) and January 2014 (wet season) Rain watersamples were collected opportunistically at Twee Rivieren North Kwang and Nossob Allwater samples were stored in 20 ml glass screw top bottles (Wheaton Liqid ScintillationVials Millville NJ USA) before analyses for both 2HH and 18O16O using a Thermo DeltaPlus XPMass Spectrometer (HamburgGermany) at theUniversity of Cape Town The samemass spectrometer was used to determine 13C12C ratios of the leaves from our study trees

Plant moisture stressLeaf stable carbon isotope ratiosWe use the stable carbon isotope ratios of leaves to determine the efficiency of carbonassimilation and plant water use (intrinsic water use efficiency) for our study trees(Dawson et al 2002 Farquhar et al 1989) Plants regulate the amount of water lost to theatmosphere by closing or opening stomata As stomata close with a decrease in availablewater there is less carbon assimilated and less discrimination against the heavier 13C isotoperesulting in more positive leaf δ13C values (Farquhar et al 1989) In the middle of the wetseason in January 2013 we collected ten fully mature whole leaves from each of our studytrees in both the Auob (twenty four trees) and the Nossob (sixteen trees) rivers Prior tomass spectrometry the leaves were oven dried at 70 C to constant weight before beingground to a fine powder using a Retsch MM200 ball mill (Retsch Haan Germany)

Xylem pressure potentialsXylem pressure potentials (XPPrsquos) are a relative indicator of the amount of water availableto the plant through a determination of the amount of tension the water column is under(Hempson February amp Verboom 2007 Scholander et al 1965) We determined middayXPPrsquos in both the wet and dry season of 2013 using a Scholander-type pressure chamber(PMS Instrument Company Corvallis OR USA) We specifically used midday rather thanpredawn XPPrsquos because of the hazards related to working in the dark with large carnivores

Shadwell and February (2017) PeerJ DOI 107717peerj2923 517

present in the area (Hempson February amp Verboom 2007) We also assumed that middayreadings in the middle of both the wet and dry seasons should sufficiently illustrate anydifferences in plant available water between trees at our study site

Specific leaf areaSpecific leaf area (the ratio of leaf area to leaf drymass) declineswith decreasing soilmoistureas leaves become smaller and heavier to reduce water loss and susceptibility to desiccation(Ackerly et al 2002 Liu amp Stuumltzel 2004) We collected ten fully mature whole leaves fromeach of our trees in both the wet and the dry season of 2013 The entire leaf including bothpetiole and rachis were then photographed against a white background before determiningleaf area using the open source software ImageJ (Abragravemoff Magalhatildees amp Ram 2004)

Canopy diebackUsing two photographs for each tree we determined the amount of canopy dieback on all ofour study trees Semi-deciduous species such as A erioloba and A haematoxylon are rarelyleafless in the dry season due to synchronised leaf fall and new leaf emergence (Sekhwela ampYates 2007) Twigs and branches with no leaves were therefore attributed to heavy browseor drought induced leaf mortality We photographed the canopy of each of our study treesin both the wet and the dry season of 2013 Each photograph was taken through an 18ndash200mm f35ndash63 HSM DC lens (Sigma Fukushima Japan) with PRO1 D UV (W) filterattached (Kenko Tokyo Japan) fixed at 52 mm F8 aperture using a Nikon D60 camera(Nikon Ayuthaya Thailand) set on a tripod (290 SeriesManfrotto Cassola Italy) 15m offthe ground The distance between camera and tree was measured from the mid-point of thetripod to the base of the tree Photographs were taken between 10h00 and 16h00 in two di-rections east to west and north to south unless obstructed when the direction was switchedthrough 180 The angle of the tripod head was adjusted using the spirit level set into thetripod so that the camera was always horizontal (relative to the ground) and tilted verticallykeeping the entire tree canopy just inside the field of view For the calibration of eachphotograph a retractable 5 m aluminium ranging rod (levelling staff Leica Geosystems StGallen Switzerland)was held vertically at the edge of the canopy in the field of view (Fig S1)

Each photograph was overlaid by a grid (50 cm boxes subdivided by 10) using AdobePhotoshop (CS5 v120 times 32 ccopy Adobe Systems Software Ltd Ireland) with living (leaf)and dead material (twigs lt 5 cm thick) noted at each grid intercept around the outer 15cm edge of the canopy (Fig S1) From the total number of intercepts (dead + live) thepercentage of canopy dieback was calculated for each of the two photographs and the resultaveraged per tree

STATISTICSThe data were split into three sets according to river and species A haematoxylon andA erioloba in the Auob and A erioloba in the Nossob

The differences between upstream and downstream for six dependent variables wereassessed Wilcoxon Sum Rank tests for leaf δ13C values linear mixed effects models forδ18O and δ2H values and midday xylem pressure potentials linear models for specific leaf

Shadwell and February (2017) PeerJ DOI 107717peerj2923 617

area and KruskalndashWallis Rank Sum tests followed by a Pairwise Wilcoxon with Bonferronicorrection for canopy dieback The models were developed using lsquoseasonrsquo (wetdry)lsquopositionrsquo (upstreamdownstream) and the interaction between these with individual treesas the random effect (the same trees were sampled in both seasons) All tests on the datawere assessed in R ccopyv312 (R Development Core Team 2014) and a value of plt 005required for significance

RESULTSWater sourceBorehole water levelInstrumentmalfunction for our piezometers affected recordings so that only those readingsof which we are certain were considered (Fig S2) Water level depth varied between 38 mto 46 m in the Auob and 49 m to 59 m in the Nossob In both the Auob and Nossob thewater table at the upstream borehole was lower than that of the downstream boreholeThis was unexpected but we speculate that this is as a result of a calcrete layer close to thesurface underlying the aquifer for the downstream boreholes in both rivers Both the Auobboreholes show a similar pattern in groundwater flux through time with a steady drop inwater level of plusmn4 m soon after the peak of the dry season (JulyAugust) and a subsequentrise of plusmn4 m matching the peak of the wet season rains (JanuaryFebruary) The Nossobdownstreamwater level showed a 6m drop twomonths after the peak of the dry season butonly a 2m rise during the peak of the following wet season with an unusually low asymptotefrom themiddle of thewet season thatmay be the result of instrumentmalfunction (Fig S2)

Oxygen and hydrogen stable isotopesMeteoric waters worldwide follow a Rayleigh distillation process that results in a linearrelationship between δ18O and δ2H termed the global meteoric water line (GMWL y =8x+10) (Craig 1961Gat 1996) For plant water source studies in arid environments theselinear relationships (essentially representing evaporation) are extremely useful as deep(non-evaporatively enriched) and shallower moisture sources (evaporatively enriched) canbe readily distinguished with deep water plotting more negative values and shallow watermore positive values (February West amp Newton 2007 West et al 2012) We constructedour own local meteoric water line (LMWL y = 506xminus675) from rain water samplescollected at Twee Rivieren in January and August and between North Kwang and Nossob inAugust 2013 (Fig 2) Rainfall at our study site is extremely seasonal and even after no rainfallfor several months and withmeanmidday temperatures between 30 C and 40 C our studytrees were not deciduous with δ18O and δ2H values indistinguishable from that ofgroundwater (Fig 2)

The results for all xylem water samples for both A erioloba and A haematoxylon in theAuob River plotted below the LMWL with wet season values similar to that of groundwater(Fig 2) In the dry season however downstream of the abstraction point xylemwater δ18Ovalues for both species are significantly different from groundwater δ18O values (Fig 2) Inthe Nossob δ18O values are similar to that of the Auob in the wet season in that xylemwaterδ18O values are indistinguishable from that of the groundwater (Fig 2) Conversely in the

Shadwell and February (2017) PeerJ DOI 107717peerj2923 717

Figure 2 Mean δ18O and δ2H values (plusmn1 SE) for xylem water of (A) Auob River Acacia haematoxylon(B) Auob River Acacia erioloba and (C) Nossob River Acacia erioloba for two seasons (wet and dry)Values are plotted relative to the local meteoric water line (y = 506xminus 675) with rain (+) and mean val-ues for borehole water (BH) included

Shadwell and February (2017) PeerJ DOI 107717peerj2923 817

Figure 3 Mean leaf δ13C values (plusmn1 SE) for (A) Auob River Acacia haematoxylon (B) Auob River Aca-cia erioloba and (C) Nossob River Acacia erioloba Different letters indicate significant differences at plt005 (Wilcoxon Rank Sum Test unpaired) between upstream and downstream

dry season δ18O values for xylem water in trees both upstream and downstream of theabstraction borehole are significantly different from groundwater values The downstreamvalues are however more positive than the upstream values

Plant moisture stressLeaf stable carbon isotope ratiosIn the Auob there were no significant differences in leaf δ13C values for either upstreamor downstream A haematoxylon or A erioloba (Fig 3) In the Nossob however A eriolobavalues downstream (minus247 plusmn 04h) were significantly more enriched (W = 52 p= 004)relative to upstream δ13C values (minus263 plusmn 037h)

Xylem pressure potentialsIn the Auob there were no significant differences in midday xylem pressure potentials(XPP) for either upstream or downstream trees (Fig 4) In the Nossob the downstreamtrees in the dry season (average minus31 MPa) had significantly more positive XPPrsquos thanupstream trees regardless of season (average minus35 and minus34 MPa respectively)

Specific leaf areaIn the Auob River both tree species downstream of the active borehole had significantlylower SLA values in the dry season than in the wet season For the Nossob River SLA valueswere overall significantly lower in the dry season than in the wet season but there was nosignificant difference between upstream and downstream (Fig 5)

Shadwell and February (2017) PeerJ DOI 107717peerj2923 917

Figure 4 Meanmidday xylem pressure potentials (plusmn1 SE) for both wet (dark) and dry (light) seasonsfor (A) Auob River Acacia haematoxylon (B) Auob River Acacia erioloba and (C) Nossob River Acaciaerioloba Different letters denote significant differences at p lt 005 (linear mixed effects model Simul-taneous tests for generalised linear hypotheses) between upstream and downstream trees within and be-tween seasons

Canopy diebackThere were no significant seasonal differences in average percentage canopy dieback forA haematoxylon in the Auob and A erioloba in the Nossob (Fig 6) In the Auob A eriolobashowed significantly more dieback downstream in both seasons Average canopy diebackwas higher in A haematoxylon (between 26 and 30) than in A erioloba (between 12 and20)

DISCUSSIONThe depth to the aquifer in the two rivers at our study site varies between 38 m and 59m with our water isotope analysis demonstrating that the trees in these rivers are usingthis deep water While ours is the first study to demonstrate the use of deep groundwaterfor A haematoxylon these results are in agreement with research 120 km south east of ourstudy site in the Kuruman River showing this for A erioloba (Schachtschneider amp February2013) Our results also show that there are differences between the low water use site inthe Auob and the high water use site in the Nossob In the Auob both of our study speciesuse deep ground water in the wet season In the dry season however these trees use deepgroundwater upstream of the borehole and in losing contact with the groundwater usemore isotopically enriched soil water downstream Significantly lower specific leaf areavalues for both species downstream indicate that in the dry season these downstream trees

Shadwell and February (2017) PeerJ DOI 107717peerj2923 1017

Figure 5 Mean specific leaf area (plusmn1 SE) for both wet (dark) and dry (light) seasons for (A) Auob RiverAcacia haematoxylon (B) Auob River Acacia erioloba and (C) Nossob River Acacia erioloba Differ-ent letters denote significant differences at plt 005 (ANOVA Tukey HSD) between upstream and down-stream within and between seasons

have physiologically adapted to the change in available water by producing smaller heavierleaves with marginally but not significantly more negative XPPrsquos

As in the Auob Aerioloba in the Nossob use deep ground water in the wet season In thedry season however all trees both upstream and downstream use more isotopically en-riched soil water and have smaller heavier leaves Significantly more positive δ13C values fordownstream trees suggest that these trees aremaintaining leaf water potential through stom-atal closure a conclusion verified by more positive XPPrsquos (McDowell et al 2008) Theseresults demonstrate that trees in the Nossob adjust to lower levels of water availability byclosing stomata to maintain relatively high water potentials thereby preventing loss of ordamage to xylem conducting tissue as is indicated in the low percentages of canopy dieback(Jones amp Sutherland 1991 West et al 2012) With only a change in SLA but no changein XPPrsquos or leaf δ13C values and little canopy dieback Aerioloba in the Auob do notdemonstrate the same levels of stress to changes in available water

At our study site in the Nossob near a borehole with a higher extraction rate than atour study site in the Auob the A erioloba trees demonstrate physiological adaption to aflux in the water table of between 45 m and 52 m between the wet and the dry seasonThis flux in water table depth is similar to the 5 m flux attributed to water abstraction inthe north-western sector of South Africa (Van Dyk et al 2008) In the Sonoran desert ofthe American South West groundwater declines of 18ndash30 m have resulted in low stem

Shadwell and February (2017) PeerJ DOI 107717peerj2923 1117

Figure 6 Mean percentage canopy dieback (plusmn1 SE) for both wet (dark) and dry (light) seasons for (A)Auob River Acacia haematoxylon (B) Auob River Acacia erioloba and (C) Nossob River Acacia eri-oloba Different letters indicate significant differences at plt 005 (KruskalndashWallis Rank Sum test followedby a Pairwise Wilcoxon with Bonferroni correction) between upstream and downstream

water potentials reduced leaf size and high levels of canopy mortality in Prosopis velutina(Stromberg 1993) Similarly in the Murray Darling Basin of SE Victoria in Australiadeclines in water level of between 121 and 226 m for Eucalyptus camaldulensis and 126and 267 m for Eucalyptus populnea resulted in significantly poorer canopy condition (Kathet al 2014) Our trees in the Nossob are exhibiting similar responses with a much smallerfluctuation in the water table

Trees in the Kalahari are the deepest rooted trees in the world and our results show thatthe reason for these deep roots is to source water at depths of between 40 and 60m (Canadellet al 1996) While the Nossob trees may be physiologically tolerant of 4ndash5 m fluctuationsin water table this could change with drought such as the current 2015ndash2016 one

CONCLUSIONThe higher abstraction for theNossob camphas resulted in the trees near that camp showingsigns that groundwater depths in the dry season have exceeded thresholds identified bystudies in Australia and America as being critical (Kath et al 2014 Stromberg 1993) Thedecline in water table in the Kalahari is not as much as that in Australia and America butthe water column in the Kalahari trees is under greater negative pressure because of thedistance to the water table of 40ndash60 m as opposed to 5 m in the American SW (Koch et

Shadwell and February (2017) PeerJ DOI 107717peerj2923 1217

al 2004 Ryan amp Yoder 1997) The increase in dry season tension on the water columnresulting from a decline in the water table has resulted in the Nossob trees controlling forwater loss through stomatal regulation However in the event of a drought such stomatalregulation can result in carbon starvation canopy dieback and canopy death (McDowell etal 2008 West et al 2012) Our perception is that there are many more dead trees in theNossob than there are in the Auob We did not evaluate this but we speculate that thesedeaths were the result of some past drought event

National Parks not only in South Africa but throughout the world are under pressureto increase revenue This growing demand for ecotourism will increase pressure ongroundwater resources in arid systems where there are no alternative water sourcesAquifer management may be described as the art of abstracting no more water than isreplenished Such abstraction should take into consideration the ecological reserve setout in the National Water Act (Republic of South Africa 1998) that allows for an adequateand timeous supply of water to maintain the integrity of not only the rivers but also theterrestrial plants reliant on the aquifer (Baron et al 2002) There has been no publishedresearch on abstraction or replenishment rates for any of the aquifers supplying the restcamps in the Park Our results would suggest very strongly that this should be a researchfocus for the future and that the Park should develop a strategic adaptive managementapproach for groundwater use Such an approach would develop thresholds for potentialconcern (Biggs amp Rogers 2003) that would allow for an adequate ecological reserve as ourstudy suggests that trees in the Park and in particular the Nossob are nearing and havepossibly exceeded the threshold of physiological tolerance

ACKNOWLEDGEMENTSWe would like to thank South African National Parks for allowing us to do the researchin the Kgalagadi Transfrontier Park We are grateful to Graeme Ellis Paola Vimercati andAmy Betzelberger for help with the fieldwork

ADDITIONAL INFORMATION AND DECLARATIONS

FundingThe research was funded by the Andrew Mellon Foundation of New York 30600716 Thefunders had no role in study design data collection and analysis decision to publish orpreparation of the manuscript

Grant DisclosuresThe following grant information was disclosed by the authorsAndrew Mellon Foundation of New York 30600716

Competing InterestsThe authors declare there are no competing interests

Shadwell and February (2017) PeerJ DOI 107717peerj2923 1317

Author Contributionsbull Eleanor Shadwell conceived and designed the experiments performed the experimentsanalyzed the data wrote the paper prepared figures andor tables reviewed drafts of thepaperbull Edmund February conceived and designed the experiments performed the experimentsanalyzed the data contributed reagentsmaterialsanalysis tools wrote the paperprepared figures andor tables reviewed drafts of the paper

Data AvailabilityThe following information was supplied regarding data availability

South African National Park Data Repository httpdataknpsanparksorgsanparksmetacatjudithk1111414sanparks

Supplemental InformationSupplemental information for this article can be found online at httpdxdoiorg107717peerj2923supplemental-information

REFERENCESAbragravemoff MDMagalhatildees PJ Ram SJ 2004 Image processing with ImageJ Biophotonics

International 1136ndash42Ackerly D Knight CWeiss S Barton K Starmer K 2002 Leaf size specific leaf area and

microhabitat distribution of chaparral woody plants contrasting patterns in specieslevel and community level analyses Oecologia 130449ndash457DOI 101007s004420100805

Baron JS Poff NL Angermeier PL DahmCN Gleick PH Hairston Jr NG JacksonRB Johnston CA Richter BD Steinman AD 2002Meeting ecological and societalneeds for freshwater Ecological Applications 121247ndash1260DOI 1018901051-0761(2002)012[1247MEASNF]20CO2

Barron O Froend R Hodgson G Ali R DawesW Davies P McFarlane D 2014Projected risks to groundwater-dependent terrestrial vegetation caused by changingclimate and groundwater abstraction in the Central Perth Basin Western AustraliaHydrological Processes 285513ndash5529 DOI 101002hyp10014

Biggs HC Rogers KH 2003 An adaptive system to link science monitoring andmanagement in practice In Biggs HC Rogers KH Du Toit J eds The Krugerexperience Ecology and management of savanna heterogeneity Washington DCIsland Press 59ndash80

Binns T Nel E 2002 Tourism as a local development strategy in South Africa TheGeographical Journal 168235ndash247 DOI 1011111475-495900051

Bothma JDP De Graaff G 1973 A habitat map of the Kalahari Gemsbok National ParkKoedoe 16181ndash188 DOI 104102koedoev16i1895

Canadell J Jackson R Ehleringer J Mooney H Sala O Schulze E-D 1996Maximumrooting depth of vegetation types at the global scale Oecologia 108583ndash595DOI 101007BF00329030

Shadwell and February (2017) PeerJ DOI 107717peerj2923 1417

Chen Y Chen Y Xu C LiW 2016 The effects of groundwater depth on water uptakeof Populus euphratica and Tamarix ramosissima in the hyperarid region of North-western China Environmental Science and Pollution Research 2317404ndash17412DOI 101007s11356-016-6914-8

Craig H 1961 Isotopic variations in meteoric waters Science 1331702ndash1703Dawson TE Ehleringer JR 1991 Streamside trees that do not use stream water Nature

350335ndash337 DOI 101038350335a0Dawson TE Mambelli S Plamboeck AH Templer PH Tu KP 2002 Stable iso-

topes in plant ecology Annual Review of Ecology and Systematics 33507ndash559DOI 101146annurevecolsys33020602095451

DeanWMilton S Jeltsch F 1999 Large trees fertile islands and birds in arid savannaJournal of Arid Environments 4161ndash78 DOI 101006jare19980455

Du Preez H Steyn G 1992 A preliminary investigation of the concentration of selectedmetals in the tissues and organs of the tigerfish (Hydrocynus vittatus) from theOlifants River Kruger National Park South AfricaWater SA 18131ndash136

Ehleringer J Dawson T 1992Water uptake by plants perspectives from stable isotopecomposition Plant Cell amp Environment 151073ndash1082DOI 101111j1365-30401992tb01657x

Farquhar G Hubick K Condon A Richards R 1989 Carbon isotope fractionation andplant water-use efficiency In Rundel P Ehleringer JR Nagy KA eds Stable isotopesin ecological research Berlin Heidelberg Springer Science amp Business Media 21ndash40

February EC Higgins SI Newton RWest AG 2007 Tree distribution on a steepenvironmental gradient in an arid savanna Journal of Biogeography 34270ndash278DOI 101111j1365-2699200601583x

February ECWest AG Newton RJ 2007 The relationship between rainfall water sourceand growth for an endangered tree Austral Ecology 32397ndash402DOI 101111j1442-9993200701711x

Gat JR 1996 Oxygen and hydrogen isotopes in the hydrologic cycle Annual Review ofEarth and Planetary Sciences 24225ndash262

Groom PK Froend RHMattiske EM 2000 Impact of groundwater abstraction on aBanksia woodland Swan Coastal Plain Western Australia Ecological Management ampRestoration 1117ndash124 DOI 101046j1442-8903200000033x

Hall CM 2001 Trends in ocean and coastal tourism the end of the last frontier Ocean ampCoastal Management 44601ndash618 DOI 101016S0964-5691(01)00071-0

Hempson GP February EC VerboomGA 2007 Determinants of savanna vegetationstructure insights from Colophospermum mopane Austral Ecology 32429ndash435DOI 101111j1442-9993200701712x

Huntley BJ Walker BH (eds) 2012 Ecology of tropical savannas New York SpringerScience amp Business Media

Jones H Sutherland R 1991 Stomatal control of xylem embolism Plant Cell ampEnvironment 14607ndash612 DOI 101111j1365-30401991tb01532x

Shadwell and February (2017) PeerJ DOI 107717peerj2923 1517

Kath J Reardon-Smith K Le Brocque A Dyer F Dafny E Fritz L BatterhamM 2014Groundwater decline and tree change in floodplain landscapes identifying non-linear threshold responses in canopy condition Global Ecology and Conservation2148ndash160 DOI 101016jgecco201409002

Koch GW Sillett SC Jennings GM Davis SD 2004 The limits to tree height Nature428851ndash854 DOI 101038nature02417

Leistner O 1959 Notes on the vegetation of the Kalahari Gemsbok National Park withspecial reference to its influence on the distribution of antelopes Koedoe 2128ndash151DOI 104102koedoev2i1860

Lite SJ Stromberg JC 2005 Surface water and ground-water thresholds for main-taining PopulusndashSalix forests San Pedro River Arizona Biological Conservation125153ndash167 DOI 101016jbiocon200501020

Liu F Stuumltzel H 2004 Biomass partitioning specific leaf area and water use efficiencyof vegetable amaranth (Amaranthus spp) in response to drought stress ScientiaHorticulturae 10215ndash27 DOI 101016jscienta200311014

Mbaiwa JE 2003 The socio-economic and environmental impacts of tourism develop-ment on the Okavango Delta north-western Botswana Journal of Arid Environments54447ndash467 DOI 101006jare20021101

McDowell N PockmanWT Allen CD Breshears DD Cobb N Kolb T Plaut J SperryJ West AWilliams DG 2008Mechanisms of plant survival and mortality duringdrought why do some plants survive while others succumb to drought NewPhytologist 178719ndash739 DOI 101111j1469-8137200802436x

Mills M Retief P 1984 The response of ungulates to rainfall along the riverbeds of thesouthern Kalahari Koedoe 27129ndash141 DOI 104102koedoev27i2574

Milton S DeanW 1995How useful is the keystone species concept and can it beapplied to Acacia erioloba in the Kalahari Desert Zeitschrift Fuer Oekologie UndNaturschutz 3147ndash156

Mucina L RutherfordM (eds) 2006 The vegetation of South Africa Lesoto and Swazi-land Pretoria South African National Biodiversity Institute 809 pp

Nel JL Roux DJ Maree G Kleynhans CJ Moolman J Reyers B Rouget M CowlingRM 2007 Rivers in peril inside and outside protected areas a systematic ap-proach to conservation assessment of river ecosystems Diversity and Distributions13341ndash352 DOI 101111j1472-4642200700308x

Neto F 2003 A new approach to sustainable tourism development moving beyondenvironmental protection Natural Resources Forum 27212ndash222DOI 1011111477-894700056

Nix H Austin M 1973Mulga a bioclimatic analysis Tropical Grasslands 79ndash21RDevelopment Core Team 2014 R a language and environment for statistical

computing Vienna R Foundation for Statistical Computing Available at httpswwwr-projectorg

Republic of South Africa 1998 Republic of South Africa National Water Act Cape TownGovernment Gazette

Shadwell and February (2017) PeerJ DOI 107717peerj2923 1617

Rogers K Biggs H 1999 Integrating indicators endpoints and value systems instrategic management of the rivers of the Kruger National Park Freshwater Biology41439ndash451 DOI 101046j1365-2427199900441x

RyanMG Yoder BJ 1997Hydraulic limits to tree height and tree growth Bioscience47235ndash242 DOI 1023071313077

SANParks 2016 Kgalagadi Transfrontier Park 2004ndash2016 Available at httpwwwsanparksorgparks kgalagadiallphp (accessed on 20 January 2016)

Schachtschneider K February EC 2013 Impact of Prosopis invasion on a keystone treespecies in the Kalahari Desert Plant Ecology 214597ndash605DOI 101007s11258-013-0192-z

Scholander PF Hammel H Bradstreet ED Hemmingsen E 1965 Sap pressure invascular plants Science 148339ndash346 DOI 101126science1483668339

Sekhwela M Yates D 2007 A phenological study of dominant acacia tree species inareas with different rainfall regimes in the Kalahari of Botswana Journal of AridEnvironments 701ndash17 DOI 101016jjaridenv200612006

Stromberg J 1993 Riparian mesquite forests a review of their ecology threats andrecovery potential Journal of the Arizona-Nevada Academy of Science 27111ndash124

Van Dyk GS Makhetha J Potgieter D Zikali T Leeme V Moletsane F Vonya T 2008Groundwater Resources in the Northern Cape Province 2008 (combined) InGroundwater status poster Kimberley Department Water Affairs and Forestry

Van RooyenMW Van Rooyen N Bothma JDP Van den Berg HM 2008 Landscapes inthe Kalahari Gemsbok National Park South Africa Koedoe 5099ndash112

VanWyk P Le Riche E 1984 The Kalahari Gemsbok National Park 1931ndash1981 Koedoe2721ndash31

West AG Dawson TE February E Midgley GF BondW Aston TL 2012 Diversefunctional responses to drought in a Mediterranean-type shrubland in South AfricaNew Phytologist 195396ndash407 DOI 101111j1469-8137201204170x

White JW Cook ER Lawrence JR 1985 The DH ratios of sap in trees implications forwater sources and tree ring DH ratios Geochimica et Cosmochimica Acta 49237ndash246DOI 1010160016-7037(85)90207-8

Shadwell and February (2017) PeerJ DOI 107717peerj2923 1717